For rotating members, in particular rotating shafts, it is sometimes necessary to receive accurate rotational information, which may be rotational position, velocity, and acceleration information. Various sensing systems have been developed to accomplish this task.
One sensing system in particular that works well in relatively harsh environments, such as with a crankshaft of an internal combustion engine, is a toothed sensor wheel. For this particular sensing system, the wheel is ferromagnetic and an inductive (magnetic field) sensor is located near the wheel periphery. As the wheel rotates, the teeth pass by the sensor, changing the magnetic field. The information is then communicated to a processor via a generally sinusoidal voltage signal from the sensor. This works generally well since it is non-contact--there are no rubbing parts to wear out, dirt and oil won't generally interfere with the signal, and the temperature effects are minimal. Generally the sensor wheel will have a series of teeth that are the same size and evenly spaced circumferentially about the wheel, with one of the teeth missing. The missing tooth location will provide a gap for indexing, to determine the absolute rotational position. This information can then be used for generally controlling engine operating parameters, such as ignition timing, fuel injector timing, etc.
While the information provided by the sensor system is sufficient for conventional internal combustion engines, the need arises to increase the accuracy of readings for this type of system in order to obtain more precise engine operation information. An example of such an instance is the desire to use a toothed crankshaft sensor wheel to detect engine cylinder misfires. It must be very precise because the slight acceleration of the crankshaft due to a cylinder firing must be determined. For this type of calculation, as little as 10 microseconds error may be too much to obtain the desired accuracy.
In general the toothed wheel sensor system produces a sinusoidal signal that has periodic zero crossings (i.e. where the voltage is zero). These zero crossings are subsequently used for determining the rotational information needed for misfire detection. The sinusoidal signal is sent to a processor for generation of a square wave from which edges are time stamped for further digital signal processing as part of a misfire monitor.
An accuracy concern arises however around the location of the missing tooth. For these inductive sensors, the missing tooth location provides for a different rate of change in magnetic flux linkage than do the other teeth on the wheel, so that residual stored energy will occur due to the loss of this flux coupling at the location of the missing tooth. The additional energy is stored in the inductor of the sensor and decays based on the particular sensor and input circuit characteristics. This residual energy will then result in higher voltages, affecting the signal for a few teeth past the gap as the excess energy decays, inherently causing a time delay in the zero crossing of the signal and hence increases the variation in the edge placement for the square waves which are subsequently generated. This, then, results in inaccurate time stamp data at these locations. The need arises then for compensation in the signal due to the energy storage in the inductive sensor.
One method of correction employed is to take the signal from the sensor as is, with the error, and employ software in a signal processor to manipulate the signal in order to compensate for the error. However, the accuracy can be less than satisfactory since the correction is based on operation at a given operating speed to minimize the software complexity, and as the rotational speed varies from the given speed, the accuracy of the error correction is reduced.
Thus, it is desirable to assure accuracy in the signal initially sent from the inductive sensor, (i.e. reduce the error at the source), and avoid the need for the error compensation in the software of the signal processor in order to obtain accurate rotational acceleration data from a sensor wheel.